What do you mean by giving results ? I think you have a supervised system, and you have your own results and you would like to learn the model according to the results you have. If that is the case, it's fine you can do that using MATLAB.
I guess you want to assess variable contributions in NNs that results conventional regression parameters: see Olden, J.D., Jackson, D.A. Illuminating the ‘‘black box’’: a randomization approach for understanding variable contributions in artificial neural networks. Ecological Modelling 154 (2002) 135–150.
NN takes the input and processess it appropriately as you have designed the network. The question here is so general. If u want to find the correct issue, you elaborate what network was chosen, transfer function, weights, hidden layer, normalization method followed etc..
Thanks for all your answers. I'm using hidden layer method. Is it good giving resgression results to NN network directly. For a proper NN network how much data sholud i give neural network??